"Knock knock! Who's there?" A study on scholarly repositories' availability

TL;DR

This study analyzes the availability of scholarly repositories listed in major registries, revealing that a significant portion of these resources are often inaccessible over time, impacting open science practices.

Contribution

It provides the first large-scale quantitative analysis of repository availability across multiple scholarly registries, highlighting link rot issues.

Findings

Over 13,000 repository URLs analyzed.

A substantial percentage of repositories are currently unavailable.

Availability varies significantly across registries.

Abstract

Scholarly repositories are the cornerstone of modern open science, and their availability is vital for enacting its practices. To this end, scholarly registries such as FAIRsharing, re3data, OpenDOAR and ROAR give them presence and visibility across different research communities, disciplines, and applications by assigning an identifier and persisting their profiles with summary metadata. Alas, like any other resource available on the Web, scholarly repositories, be they tailored for literature, software or data, are quite dynamic and can be frequently changed, moved, merged or discontinued. Therefore, their references are prone to link rot over time, and their availability often boils down to whether the homepage URLs indicated in authoritative repository profiles within scholarly registries respond or not. For this study, we harvested the content of four prominent scholarly registries and resolved over 13 thousand unique repository URLs. By performing a quantitative analysis on such an extensive collection of repositories, this paper aims to provide a global snapshot of their availability, which bewilderingly is far from granted.